基于MAS协作机制的动态计算资源调度模型

针对网格环境的自治性、动态性、分布性和异构性等特征.提出基于多智能体系统(Mutil Agent System,MAS)博弈协作的资源动态分配和任务调度模型,建立了能够反映供求关系的网格资源调度模型和任务求解算法,证明了资源分配博弈中Nash均衡点的存在性、唯一性和Nash均衡解,该方法能够利用消费者agent的学习和协商能力,考虑和引入消费者的心理行为,使得消费者的资源申请和任务调度具有较高的合理性和有效性.实验结果表明,资源调度算法不但可以有效减少不必要的延迟,而且在响应时间的平滑性、吞吐率及资源利用率方面比传统算法要好,从而使得整个资源的供需合理、负载均衡.

MAS Coordination Mechanisms Based on Grid Resource Scheduling Model

Autonomy for the grid environment,dynamic,distributed and heterogeneous and other characters.Proposed based on multi-agent system(Mutil Agent System,MAS) Dynamic Resource Allocation Game collaboration and task scheduling model,a supply and demand can Fanying Grid resources and task scheduling model solving algorithm,proved resource allocation game in the Nash equilibrium existence,uniqueness and the Nash equilibrium solution,the method can take advantage of the consumer agent's ability to learn and consultations,to consider and the introduction of consumer psychological behavior,making consumers of resources and task scheduling for high rational and effective.Experimental results show that resource allocation algorithm can not only reduce unnecessary delays,and the smoothness of the response time,throughput and resource utilization better than the traditional algorithm,which makes the whole supply and demand of resources and reasonable load balancing.